Ask 100 random people in the US whether they think "top" is "better associated" with "good" and "down" is "better associated" with "bad", or the other way round. You can even use arrows and randomize the way you ask the question, if you want.

If you come up with a majority of people telling you "down" is "better associated" with "good", I'll live stream myself on Twitch eating the pair of socks I'm currently wearing.

Also, how typical HN to take something that's absolutely obvious and deny it, just so you can escape the terrible idea that you might be subject to unconscious bias.

I think you're addressing the wrong part of the argument. Of course there are loose associations between concepts that manifest on abstract word-association tasks.

It is a considerably stronger yet less-supported statement that these biases fundamentally corrupt your thinking: that you look at Australia and can't help yourself but think it's 10% worse than Greenland.

It is an even stronger and even less-supported statement the world is going to be better off if we stop using certain tainted words or drawing maps in a certain way - i.e., that these biases hurt people and can be excised with one simple linguistic or cartographic trick.

It's a lot easier to interpret these debates as the manifestation of a bad personality trait: the desire to get sanctimonious about how other people are living their lives.